Altered baseline immunological state and impaired immune response to SARS-CoV-2 mRNA vaccination in lung transplant recipients

Abstract

The effectiveness of COVID-19 mRNA vaccines is diminished in organ transplant patients. Using a multi-omics approach, we investigate the immunological state of lung transplant (LTX) recipients at baseline and after SARS-CoV-2 mRNA vaccination compared to healthy controls (HCs). LTX patients exhibit a baseline immune profile resembling severe COVID-19 and sepsis, characterized by elevated pro-inflammatory cytokines (e.g., EN-RAGE [also known as S100A12], interleukin [IL]-6), reduced human leukocyte antigen (HLA)-DR expression on monocytes and dendritic cells, impaired cytokine production, and increased plasma microbial products. Single-cell RNA sequencing identifies an enriched monocyte cluster in LTX patients marked by high S100A family expression and reduced cytokine and antigen presentation genes. Post vaccination, LTX patients show diminished antibody, B cell, and T cell responses, along with blunted innate immune signatures. Integrative analysis links these altered baseline immunological features to impaired vaccine responses. These findings provide critical insights into the immunosuppressed condition of LTX recipients and their reduced vaccine-induced adaptive and innate immune responses.

Keywords: endotoxin tolerance; human immunology; immunocompromised; lung transplant; monocytes; multi-omics; sepsis; systems immunology; systems vaccinology; vaccines.

Graphical abstract

Figure 1

Humoral immune response in LTX recipients (A) Schematic of the study design and participants’ details. The schematic was made using BioRender. (B) Antibody binding titers to WT strain in longitudinal samples from HC (n = 21) and LTX patients (n = 19). The dotted line indicates the cutoff value. The numbers are geometric mean titers. (C) Cross-binding breadth against other SARS-CoV-2 variants, which represents the ratio of anti-VOC spike binding titers to anti-Wuhan spike titers. Error bars represent the 95% confidence interval (CI). (D) Live virus-neutralizing antibody titers against the WA.1 and XBB.1 strains. The dotted line indicates the cutoff value. The numbers are geometric mean titers. (E) Spearman correlation between binding or neutralizing titer of LTX patients and the number of days since transplantation until the first dose of mRNA vaccine. 95% confidence interval is colored in gray. (F) Spike protein concentrations in the longitudinal plasma samples. (G) Spearman correlations between the concentrations of plasma spike protein at day 1 and the binding Ab titers at pre-boost, shown in all participants combined and the LTX patient group independently. 95% confidence interval is colored in gray. Statistical significance between time points was conducted with Wilcoxon matched-pairs signed-rank test. Comparisons between the groups were conducted with Mann-Whitney test.

Figure 2

Plasma cytokine response to mRNA vaccine in LTX patients (A) Principal component analysis (PCA) analysis of plasma protein profiles of the HC and LTX groups at all time points (HC: n = 21, LTX: n = 19). (B) Bar plots showing the significantly differentially expressed proteins between the groups. Shown are adjusted p < 0.05 and FC > 2. Proteins preferentially produced in LTX patients shown in red. Dot plots show the comparison of EN-RAGE, IL-8, CASP-8, and IL-10 production. The dotted line indicates the cutoff value. Comparisons between the groups were conducted with Welch’s t test after testing for normal distribution. (C) Volcano plot showing the fold change of plasma proteins at day 1, and day 7 relative to the baseline (HC: n = 21, LTX: n = 19). (D) Comparison of the levels of key proteins IFN-γ, CXCL10, MCP-3, and IL-6 at baseline and following vaccination between the two cohorts. The dotted line indicates the cutoff value. Comparison between groups and time points was conducted with two-way ANOVA with mixed effects. (E and F) Spearman correlation of baseline IL-8 concentration with neutralizing antibody titer (left) or with binding antibody titer (right) at day 21 (E) and day 90 (F), respectively. Line represents liner regression. (G) Spearman correlation of baseline CASP-8 concentration with binding antibody titer at day 21.

Figure 3

Adaptive immune responses to mRNA vaccine in LTX patients (A) Flow chart showing the gating strategy for the spike-specific memory B cells (HC: n = 21, LTX: n = 19). (B and C) Frequency of SARS-CoV-2 (B) WA.1 or (C) Omicron BA.5 spike-specific IgG⁺ memory B cells calculated as the proportion out of total CD20⁺ B cells. Dashed line plots on the left show each individual donor, and the plot on the right shows the geometric mean value from each group and shaded error bar with a confidence interval of 95%. (D) Pie chart with the shaded area representing the proportion of WA.1 and BA.5 cross-binders out of total WA.1 binding memory B cells. Blank areas show the proportion of memory B cells binding only to the WA.1 spike. HC is shown in blue circles, and LTX is shown in red circles. (E and F) The frequency of ancestral spike-specific CD4⁺ T cells secreting (E) IL-2, IFN-γ, or TNF (Th1-type), IFN-γ alone, (F) IL-4 (Th2 type), IL-21, and CD40L (cTfh type) out of total CD4⁺ T cells. Median responses with 95% confidence interval (CI) are plotted (HC: n = 21, LTX: n = 19). (G) Pie charts showing the proportion of spike-specific CD4⁺ T cells producing 1, 2 (blue arc), or 3 cytokines (black arc) at respective time points. The proportion of IFN-γ producers is highlighted as the red arc. (H and I) Spearman correlation of T cell responses at pre-boost or at day 7, respectively, with (H) day 21 binding Ab titer and (I) day 90 neutralizing titer. Statistical significance between time points was conducted with Wilcoxon matched-pairs signed-rank test. Comparisons between the groups were conducted with Mann-Whitney test.

Figure 4

LTX patients exhibit distinct baseline transcriptome alterations and an attenuated transcriptional response characterized by a remaining IFN antiviral signature (A) The number of differentially expressed genes (DEGs) (absolute log2-transformed fold change >0.2 and Wald p < 0.01) between two groups at pre-boost time point. Shown is LTX versus HC (LTX, n = 19; HC, n = 21). (B) PCA analysis of all genes from both groups shown at all time points. (C) BTMs that are significantly different between LTX vs. HC group at pre-boost (false discovery rate [FDR] < 0.05, absolute normalized enrichment score [NES] > 2). (D) Number of DEGs (absolute log2-transformed fold change >0.2 and Wald p < 0.01) at day 1 and day 7 relative to the pre-boost time point. Size-proportional Venn diagram is plotted to visualize the proportion of shared and unique regulated genes on day 1. (E) The enrichment of BTMs at day 1 and day 7 (FDR < 0.05, absolute NES > 2). BTMs were grouped by functions and were color-coded as indicated. (F and G) Line graphs depicting fold changes in the enrichment of (F) antigen presentation (II) (M95.0) and monocyte (II) (M11.0) modules, and (G) antiviral IFN signature (M75) module relative to pre-boost. HCs and LTX groups are compared side by side. Lines represent the genes in each respective module.

Figure 5

Baseline immune disparities between LTX patients and HC at the single-cell level (A) Uniform manifold approximation and projection (UMAP) representation of 17 (C0–C16) cell clusters identified by single-cell transcriptional profiling (HC, n = 9; LTX, n = 5). (B) UMAP showing LTX patients and HC separately at pre-boost time point, with arrows highlighting the key features. (C) Stacked bar plot illustrating the proportion of each cell cluster relative to the total number of cells. Each bar represents an individual participant, and cell clusters are categorized into myeloid cells and lymphoid cells. (D) Volcano plot comparing cellular frequencies for each cell cluster between HCs and LTX patients. (E) Gene Ontology (GO) analysis of all differentially expressed genes for the corresponding classical monocyte clusters at pre-boost time point. (F) Heatmap displaying the top 50 most differentially expressed genes from each of the four clusters at the baseline. (G) Heatmap presenting the top differentially expressed genes from direct baseline comparison of the two groups across each cluster. Color indicates the natural log of fold change (LTX vs. HC). (H) Expression of S100A12, S100A8, S100A9, and FKBP5 from bulk RNA-seq data. HC, n = 21, and LTX, n = 19, with p values calculated by Mann-Whitney U test. (I) Spearman correlation of S100A12 expression in major cellular sources with EN-RAGE levels in plasma at the pre-boost time point. (J) HLA-DR protein expression on the surface of myeloid cells assessed by single-cell antibody sequencing, presented at each participant level. p values were calculated with Mann-Whitney U test.

Figure 6

Single-cell transcriptional response in LTX patients reveals a notable absence of induction in ISG hi monocytes (A) UMAP plotted based on time points and groups, with arrows highlighting the key features. Each UMAP plot represents an equal number of cells. (B) Spearman correlation between day 1 IFN-γ concentration and the day 1 frequency of C8 (CD14_Mono [ISG hi]). Each dot represents one participant (HC, n = 9; LTX, n = 5). p values are calculated for all participants in black text and for HC group in blue text. (C) ISG score displayed in UMAP based on time points and groups. (D) The GSEA of top-enriched BTMs at day 1 for each individual cluster (false discovery rate [FDR] < 0.05). The color of the circle represents the normalized enrichment score, and the size of the circle represents the percentage of genes from each respective BTM that are leading genes.

Figure 7

Innate immune cells exhibit impaired cytokine secretion and a similarity toward patients with severe COVID-19 (A and B) Boxplots (A) and contour plots (B) displaying the fraction of mDCs in PBMCs from HC or LTX patients producing TNF-α, IL-1β, or IL-6 following stimulation with the viral or bacterial cocktail (HC, n = 9; LTX, n = 5). Bars show the range of all data points (minimum to maximum). (C and D) Boxplots (C) and representative contour plots (D) depicting the fraction of monocytes in PBMCs from HC or LTX patients producing TNF-α, TNF-α⁺IL-6⁺, or TNF-α⁺IL-1β⁺ in response to stimulation with the viral or bacterial cocktail. (E) Boxplots showing the fraction of pDCs in PBMCs from HC or LTX patients producing IFN-α or TNF-α upon viral cocktail stimulation. (F) Fold change of NF-κβ p65 (Ser529) staining in mDCs stimulated with bacterial cocktail relative to no stimulation in HC and LTX patients. (G) LPS concentrations measured in the plasma of HC or LTX patients at pre-boost time points. Black dots represent donors whose samples were subjected to single-cell RNA sequencing. (H) MMRN analysis integrates datasets from baseline Olink and bulk RNA-seq, day 21 and day 90 binding, neutralizing antibody titers day 7 Th1 cytokine-producing CD4⁺ T cells, and day 21 memory B cells frequencies (HC: n = 21, LTX: n = 19). Line thickness indicates Spearman’s rank correlation coefficient. Red and blue lines denote positive and negative correlations with vaccine responses, respectively. The differential expression of each community between LTX patients and HC is highlighted with either red or blue borders around the label. (I) Heatmap showing the mean differences of average module score between LTX patients and HC across all cell clusters in single-cell data at the pre-boost time point. Color indicates the value of the Mean Diff, and the size of the square shows the −log₁₀(adjusted p value). p values were calculated with Mann-Whitney U test.